True rack-level heat and electrical replication for end-of-line testing, with solid-state switching, before a single GPU is installed.
AI has changed the behaviour of the load, not just its size. When it rises or drops suddenly, the CDU, pumps, valves, sensors and controls have to respond, then the cooling plant, UPS and generators behind them. A steady load bank cannot show you that.
Solid-state switching steps the load far faster than contactor-based banks, in both directions.
The test covers every part of the cooling and power chain, not only the capacity of the loop.
Not every part of a server is liquid cooled, so the ACLB 60 recreates the air-cooled heat as well.
Power quality and cooling are proven at the point the GPUs will run.
AI liquid GPU emulator · LCLB 160
160kWRack format
True rack-level heat and electrical replication of a liquid-cooled AI rack. Supports GB200 and GB300.
Enquire Flagship
AI liquid GPU emulator · LCLB 240
240kWPer unit, rack format
The same rack format at 240kW, for the highest-density AI racks. Supports GB200 and GB300, and is Rubin ready.
Enquire Low delta T
Air emulator · ACLB 60
60kWRack format
A low delta T air heat emulator, designed for 10 to 16°C, so the air side behaves like a real rack.
See the specificationWork out how many emulators a hall needs with the sizing tool. Testing the cooling plant while the hall is still being built? See the base build range.
Every emulator is available to buy or to hire, sized to the test plan.
The LCLB 160 and LCLB 240 both support GB200 and GB300, and the LCLB 240 is Rubin ready. Give us your design and IT load and we will size the emulation to match.
A liquid-cooled load bank that replicates how a GPU rack will behave, for heat and for electrical load, at the rack. The load moves quickly in both directions, so commissioning teams can see whether the CDU, pumps, valves, controls, cooling plant, UPS and generators respond as they should before the GPUs are installed.
A base build load bank gives a dependable, controllable heat load that proves capacity. A GPU emulator adds solid-state switching so the load moves the way real compute moves, up and down, and the whole chain has to respond.
Yes. The emulator sits at the rack position instead of at the building entry, so power and cooling are tested where the GPUs will run.
Yes. The AI liquid GPU emulators are available to hire. Speak to the team about availability for your project.
Electrical and hydraulic specification, control detail, dimensions and connections, as a PDF with no email gate.
Rack-level GPU simulator load bank. Electrical and hydraulic specification, control detail, dimensions and connections.
Download GPU Emulator 240240kW at rack level, with solid-state switching for fast, repeatable load profiling against the highest-density AI racks.
Download ACLB 60Low delta T air heat emulator in rack format, run alongside the liquid emulators to replicate the heat the liquid loop does not carry.
DownloadGive us your design and IT load, and we replicate its live performance before it goes live.